I am an undergraduate student at BITS Pilani pursuing a dual degree in B.E. (Hons.) Electrical Engineering and M.Sc. (Hons.) Mathematics.
I worked on building a surge pricing model in order to shape demand based on customer traffic and rider supply. All the data was collected via Mixpanel and Redshift. Detailed studies were done to ensure proper functioning of the model in different levels of stress and different locations.
The internship was to collect and rank YouTube videos for different brands on the basis of brand keywords. A relevance score metric was created in order to rank the videos collected via YouTube API. Google Ads API and NLP models were used to generate brand keywords and reduce noise.
Most of the data from IACTs currently in operation comes in ROOT format. Enabling ROOT input in CTLearn will open its doors to existing IACT facilities. Hence, this project aims to use CTLearn to train classification models that tell gamma-ray images from cosmic-ray images apart and why this is important in the context of IACTs, then, enable the input of data in ROOT format and train some of the already existing deep learning models in CTLearn with data from current-generation IACTs.
The internship at MRC aimed at using AIS data from vessels, and apply Deep Learning methods to predict the trajectory of the vessels. Also, using the AIS data, speed based Anomaly Detection was performed by using Machine Learning. Also, a small study was proposed on studying the effect of high density trade routes in Indian Inland waterways on the endangered Ganges river dolphin species.
Gamma-ray events collected via telescopes need a lot of cleaning to remove other cosmic-ray events from the background in order to aid event reconstruction. This work was my Master's thesis, wherein I added a separate Dense-layers based model into the existing deep learning models using Hillas parameters as inputs.
The Indian SWAN, a potential SKA-precursor, aims not only to significantly enhance Indian observing capabilities in radio but importantly, also to sustainably build & nurture future generations of talented radio astronomers in India to take up the challenges and lead in exciting research in astronomy, including those with the SKA.
Rainy weather greatly affects the visibility of salient objects and scenes in the captured images and videos. The object/scene visibility varies with the type of raindrops, i.e., adherent rain droplets on the camera lens, streaks, rain, and mist, etc. Moreover, they pose multifaceted challenges to detect and remove the raindrops to reconstruct the rain-free image. Recently, both CNN and GAN based models have been designed to remove rain droplets from a single image. In this project, we design a simple yet effective GAN framework to achieve improved deraining performance over the existing state-of-the-art methods.
The project aimed at applying Machine Learning methods to predict and classify solar flares and ionospheric disturbances. The predictions were based on data provided by SWPC GOES X-Ray Flux. Data collection is automated from loop antenna using SID software and Python.
The project aimed to be a stepping stone towards making a universal classifier for Spam SMS. A Naive-Bayes classifier was implemented with and CountVectorizer on the SMS Spam dataset. A new dataset was created by mixing 2 large English and Hindi datasets courtesy of UCI and IIITD respectively.